1. Field of the Invention
The present invention relates to image processing for detecting a specified pattern in an image.
2. Description of the Prior Art
Recently, a color copying machine provides many functions with improved performance, and counterfeit on paper money, securities or the like becomes a big problem. Then, effective countermeasures for preventing counterfeit have been developed. In one of the countermeasures, a specified pattern is embedded in a design of a paper money or the like. When an image scanned is read in copying operation in a copying machine, the scanned image is analyzed. If the specified pattern is detected in the image, normal image forming is forbidden so as not to copy the image.
When a paper money or the like is scanned, the specified pattern has to be detected from the scanned image even when the paper money or the like is put in an arbitrary position on a scanner or even when it has stains, damages or the like locally.
Most of the input data received from an input device such as a scanner are color image having many data. Further, there is a tendency that the input and output devices can deal an image at higher speed and at higher resolution. However, image processing or detection of the specified patterns is required to be performed real time in order to prevent counterfeit. Then, it is an important problem to develop a technique of image detection at high speed even in the presence of noises.
A pattern having three or more elements arranged at predetermined positions is used as one of the specified patterns for preventing counterfeit. Previously, when such a specified pattern is detected, a reference bit pattern on the arrangement of the elements is stored for comparison with an image to be checked. In order to deal with rotation of the image, a plurality of reference bit patterns displaced by a certain angle successively are prepared further. First, elements are detected in input image data. Next, the arrangement of the detected elements is compared on all the reference bit patterns. The largest degree of matching obtained by the comparison is adopted as the degree of matching.
However, this recognition technique has following problems. A large amount of reference bit patterns have to be prepared according to a plurality of rotation angles. Further, comparison is performed for the plurality of angles, efficiency of image recognition is not good. Further, the precision of image recognition becomes worse when the image is expanded or reduced.